Temporal variation in the operational sex ratio and male mating behaviours in reindeer (Rangifer tarandus

Temporal variation in the operational sex ratio and male mating behaviours

1

in reindeer (Rangifer tarandus)

2

Robert B. Weladji ^a*, Guillaume Body ^a, Øystein Holand ^b, Xiuxiang Meng^c and Mauri 3

Nieminen^d 4

5

a Department of Biology, Concordia University, 7141 Sherbrooke Street West, Montreal, QC, 6

H4B 1R6, Canada 7

b Department of Animal and Aquacultural Sciences, Norwegian University of Life Sciences, P.O.

8

Box 5003, 1432 Ås, Norway 9

c School of Environment and Natural Resources, Renmin University of China, 59 Zhongguancun 10

Ave, Beijing 100872, China 11

d Natural Resources Institute of Finland, Reindeer Research Station, 99910 Kaamanen, Finland 12

13

* Corresponding author: robert.weladji@concordia.ca

Department of Biology, Concordia University, 7141 Sherbrooke Street West, Montreal, QC H4B 14

1R6, Canada 15

Phone: 514-848-2424 ext. 3408 16

Fax: 514-848-2881 17

18 19

Running head: Variation in males mating tactics with OSR 20

ABSTRACT 21

In polygynous species, sexual selection is mostly driven by male ability to monopolize access to 22

females in oestrous. In ungulates, the operational sex ratio (OSR), i.e. the proportion of males to 23

individuals ready to mate, varies throughout the peak rut, resulting from the temporal variation in 24

the number of females in oestrous. But the way males adjust their mating tactics to maximise 25

their access to females in oestrous (i.e. as OSR varies) is yet to be investigated. Using 15 years of 26

behavioural observations in reindeer (Rangifer tarandus), we compared the relative importance 27

of time within the rutting season (days to the peak-rut) and the OSR to explain the variation in 28

the propensity (i.e. the frequency after controlling for the potential number of encounters) of 29

young and adult dominant males to engage in four mating tactics: herding females, chasing other 30

males, investigating female reproductive status, and courting females. Male-male agonistic 31

behaviour was the most frequent mating behaviour, followed by herding. As predicted, dominant 32

male mating tactics changed over the rutting season: first herding, then chasing other males, and 33

finally investigating and courting females. In contrast to our prediction, we did not find support 34

for the OSR theory. We noted some discrepancies in how young and adult dominant males 35

adjusted their tactics during the mating season, adults being more efficient in timing and in 36

performing their behaviour to maximise access to females in oestrous. The reported sequence of 37

mating tactics may be more efficient than a static mating tactic to monopolize females in 38

oestrous, regardless of the population composition.

39 40

Keywords: courtship, intrasexual aggression, mating tactics, OSR, polygyny, ungulates 41

42

1. Introduction 43

Sexual selection, the driver of the evolution of adaptations that increase the mating success of 44

certain individuals over others of the same sex and species, arises primarily from male-male 45

competition for access to mates and from female mate choice (Darwin, 1871). In polygynous 46

mammalian species, sexual selection is mostly driven by male ability to monopolize access to 47

females in oestrous (Emlen and Oring, 1977). Accordingly, male mating tactics vary with the 48

temporal and spatial distribution of females in oestrous, as well as male ability to control female 49

movement (Clutton-Brock, 1989). Classical mating systems theory predicts that a male’s ability 50

to monopolize females in oestrous, and therefore the strength of sexual selection, increases with 51

the level of competition, best measured by the operational sex ratio (OSR), i.e. the proportion of 52

males to the total number of individuals ready to mate (de Jong et al., 2012). However, male 53

ability to monopolize females in oestrous may depend on how mates are acquired (Klug et al., 54

2010). To increase their ability to monopolize females, dominant males may devote more time 55

and energy into mating behaviours, especially when the competition is stronger i.e. higher OSR 56

(Emlen and Oring, 1977), but also when there are more females to defend (lower OSR); which 57

will in turn influence their mating success (Coltman et al., 1998; Pelletier and Festa-Bianchet, 58

2006; Willisch et al., 2012).

59

When female oestrus is short and highly synchronous, such as in ungulates (de Vos et al., 60

1967), the number of females in oestrous is expected to follow an inverse U-shaped curve, with 61

many females in oestrous during the peak-rut period, and few during the early and late rut 62

periods (Hirotani, 1989). Accordingly, and as the number of mature males remains constant 63

within a single rut season in closed populations, the OSR will exhibit a U-shaped pattern, with a 64

minimum during the peak rut. OSR theory would therefore predict a U-shape curve of male 65

investment in competitive behaviour over the rut: low aggression during the peak rut and higher 66

aggression early and late in the season.

67

An alternative to the prediction made from the OSR theory is that males adjust their tactics 68

according to time in the rutting season (early, peak, and late rut), independently of the level of 69

competition. The plasticity of ungulate male mating tactic is well documented (e.g., Carranza et 70

al. 1995; Pelletier, 2005) among species, populations, years and individuals (de Vos et al., 1967;

71

Carranza, 2000; Brockmann, 2001; Mysterud et al., 2004; Isvaran, 2005). Male ungulates adjust 72

their reproductive effort to the phenology of females in oestrous (Mysterud et al., 2008) and we 73

can therefore expect them to also adjust their mating tactic, especially in regards to their 74

influence on their reproductive success. Behaviours which have an indirect benefit (e.g. herding 75

– pursuing a female until she returns to the mating group; or male-male aggressions – either 76

chasing other males from the mating group or fighting to maintain the dominance) are useless 77

toward the end of the rut, while those which have an immediate benefit (such as investigating 78

females – to assess their reproductive status and find the female currently in oestrous; or 79

courting females – following a female while performing mating displays) are useless at the 80

beginning of the rut.

81

We used 15 years of rutting behaviour data to study the phenology of male mating tactics and 82

their variation with OSR in reindeer Rangifer tarandus. Reindeer has a short mating season with 83

most females copulating within 10 days (Kojola, 1986; Skogland, 1989) and females have a short 84

oestrus (Espmark, 1964; Hirotani, 1989; Ropstad, 2000), inducing a strong temporal variation of 85

the OSR. Male reindeer mating tactics have been suggested to be particularly flexible (Clutton- 86

Brock, 1989), and males adjust their reproductive effort to local conditions, such as group size 87

and number of competitors (Tennenhouse et al., 2011). Male age has a strong influence on the 88

timing of reproductive effort (Mysterud et al., 2004; Tennenhouse et al., 2012) and also 89

influences the efficiency of male mating behaviours (L'Italien et al., 2012; Body et al., 2014).

90

Accordingly, we tested the following three predictions, the first one being associated to the 91

phenological hypothesis, the second being associated to the OSR hypothesis, and the third one 92

related to the influence of age on the reported patterns: (1) Dominant male mating tactics will 93

change with the time during the rutting season, in the following order; (a) herding females at the 94

beginning of the rut, (b) investigate and copulate with females mostly during the peak-rut and 95

then (c) court females at the end of the peak rut. We also expect inter-male agonistic behaviours 96

to increase during the peak rut. (2) Males will spend more time into each of these mating 97

behaviours with an increase in OSR, particularly for the inter-male agonistic behaviours. (3) We 98

further predicted that the expected pattern will be more pronounced for adult dominant males as 99

compared to juvenile, less experienced dominant males.

100 101

2. Methods 102

2.1. Study area and study population 103

The study was conducted at the Kutuharju Field Reindeer Research Station, in Kaamanen, 104

Finland (69°N, 27°E). We collected data from a semi-domestic Reindeer population free ranging 105

in two large fenced areas: the southeast Sinioivi (13.4 km²) and the northwest Laulavaara (13.8 106

km²). Birch Betula spp and Pine Pinus sylvestris forests, boggy areas and lakes characterized the 107

enclosures. The herd composition (a herd is the population in an enclosure in a particular year) 108

was experimentally modified every year for 15 years (1996 to 2011 except 1998) for a total of 16 109

enclosure-years (Table 1). We changed the number of males and females, and therefore the adult 110

sex ratio, as well as the male age structure, i.e. only young, only adult or mixed age structure 111

(Table 1). Apart from these experimental herd compositions, animals were free ranging within 112

enclosure limits and behaved naturally. Males were fitted with VHF radio collars while females 113

were fitted with coloured collars, both with unique identification facilitating mating group 114

composition determination and the monitoring of individual behaviour. Using Lent (1965)’s 115

definition of a group, a mating group (also called harem) was considered “an aggregation of 116

individuals separated by some distance from other aggregations, showing coordination of 117

activities, such as travelling together or resting and feeding together”, with at least one male and 118

one female (Uccheddu et al. 2015). Because individuals had ear tags, we could track their 119

identity through years (34% of the males were present two or more years). Every day from mid- 120

September to mid-October we located collared males and their harem using ground tracking, and 121

recorded group composition (number of males and females and their identities) and behaviours 122

of dominant males, i.e. harem holders which are easily identified in Rangifer. Indeed, every time 123

we found a group the dominant male was clearly recognised, occupying a central position, 124

contrary to the satellites, and performing mating behaviours more than any other male (typically 125

chasing other males, grunting, or herding females; see Tennenhouse et al. 2011 for details on 126

dominant males determination) and independently of their age.

127 128

2.2. The operational sex ratio (OSR) 129

We defined the OSR as the proportion of males to the total number of individuals ready to 130

mate, i.e. mature males and females in oestrous (de Jong et al., 2012). We calculated the OSR on 131

a daily basis at the herd level (OSR herdday) and at the group level (OSRgroup). The number of 132

males ready to mate is defined as the number of mature males in the herd or as the number of 133

mature males in a given group. We estimated the number of females in oestrous in the herd or in 134

a given group on a daily basis using a backdating procedure from birth date and three calculation 135

steps as presented below, assuming that females were in oestrous for a single day. Oestrus 136

duration has been estimated to last between 24 h and 48 h in reindeer (Espmark, 1964; Hirotani, 137

1989; Ropstad, 2000).

138

First, we estimated the mating day of every female that gave birth in each herd. We removed 139

from their birth date the gestation duration controlled for the age of the female, the sex of the calf 140

and the mating time (Eq. 1, Mysterud et al., 2009; coefficients were provided by Atle Mysterud, 141

personal communication). For further analyses, we excluded very late mating dates, i.e. which 142

occurred in November or later, as they may more likely represent a second oestrus cycle.

143

Equation 1 144

𝑀𝑎𝑡𝑖𝑛𝑔 𝑑𝑎𝑡𝑒 = 𝐵𝑖𝑟𝑡ℎ 𝑑𝑎𝑡𝑒 − 282.83 − 1.65 × 𝑆𝑒𝑥 − 0.31 × 𝐴𝑔𝑒 + 365

−0.23 + 1

Where Mating date and Birth date are in Julian days (January first = 1); Sex is calf sex (Male = 145

1; Female = 0); Age is the age of the mother when she gave birth.

146 147

Second, we estimated the statistical density of females in oestrous from the histogram 148

distribution of mating days in each herd separately. Then, we multiplied this density by the 149

number of females in the herd to obtain the expected value of the number of females in oestrous 150

in a herd at a given date (Oestrous herdday). We calculated the number of females in oestrous in a 151

group at a given date (Oestrous groupi) based on the proportion of the mature females of the herd 152

present in the group (Equation 2).

153

Equation 2 154

𝑂𝑒𝑠𝑡𝑟𝑜𝑢𝑠 𝑔𝑟𝑜𝑢𝑝_𝑖 = 𝑂𝑒𝑠𝑡𝑟𝑜𝑢𝑠 ℎ𝑒𝑟𝑑_𝑑𝑎𝑦×𝑓𝑒𝑚𝑎𝑙𝑒𝑠 𝑔𝑟𝑜𝑢𝑝_𝑖 𝑓𝑒𝑚𝑎𝑙𝑒𝑠_{ℎ𝑒𝑟𝑑}

Where Oestrous groupi and Oestrous herdday is the number of females in a given group i or on a 155

given day in the herd, respectively; females groupi and femalesherd are the number of females in a 156

given group i and in the herd, respectively.

157

158

By doing so, we made two assumptions. First, we assumed that unmated or females that 159

aborted had a similar temporal distribution of their oestrus as compared to females that gave 160

birth. Second, we assumed females in oestrous were equally distributed among mating groups.

161

Although these assumptions may be violated as youngest females are the least likely to give birth 162

and mate later (Eloranta and Nieminen, 1986; Skogland, 1989), and as females in oestrous may 163

group around particular males more than anoestrous females, i.e. female mate choice, it is the 164

most parsimonious assumption to estimate oestrus day of females that did not give birth and their 165

distribution among groups.

166

Third, we calculated the OSR as the proportion of mature males to the total number of 167

individuals ready to mate (i.e. mature males + females in oestrous), daily at the herd level 168

(Equation 3), and for each group (Equation 4). We calculated the operational sex ratio at the herd 169

level on a daily basis (OSRherd) and the operational sex ratio at the group level (OSRgroup).

170 171

Equation 3 172

𝑂𝑆𝑅 ℎ𝑒𝑟𝑑_𝑑𝑎𝑦 = 𝑚𝑎𝑙𝑒𝑠_{ℎ𝑒𝑟𝑑}

𝑚𝑎𝑙𝑒𝑠_{ℎ𝑒𝑟𝑑}+ 𝑂𝑒𝑠𝑡𝑟𝑜𝑢𝑠 ℎ𝑒𝑟𝑑_𝑑𝑎𝑦 Equation 4

173

𝑂𝑆𝑅 𝑔𝑟𝑜𝑢𝑝_𝑖 = 𝑚𝑎𝑙𝑒𝑠 𝑔𝑟𝑜𝑢𝑝_𝑖

𝑚𝑎𝑙𝑒𝑠 𝑔𝑟𝑜𝑢𝑝_𝑖 + 𝑂𝑒𝑠𝑡𝑟𝑜𝑢𝑠 𝑔𝑟𝑜𝑢𝑝_𝑖

Where OSR herdday and OSR groupi are the operational sex ratio in the herd a given day and in a 174

given group, respectively; malesherd and males groupi the number of males in the herd and in a 175

given group, respectively; Oestrous herdday and Oestrous groupi the number of females in 176

oestrous in the herd a given day or in a given group, respectively.

177 178

2.3. Timing of the mating season 179

To compare mating seasons, we centered each one on its median mate date (defined as Julian 180

Day: JD = 0). The peak-rut week was defined as the week surrounding this date and only used 181

for descriptive purpose. We centered OSR values as well as behavioural records. We analyzed 182

the data recorded during the month surrounding the median mate date (i.e. from JD = -14 to JD = 183

14) as the probability a female was in oestrous was too low before that period and to avoid an 184

overlap with a potential second peak-rut, as female reindeer can re-ovulate if they were not 185

fertilized in their first oestrus. We also reported every copulation observed while in the field.

186

These records were centered as described above, and we only displayed those who are in the 187

time interval of interest.

188 189

2.4. Dominant male mating tactics 190

Dominant male mating behaviour was observed based on the focal observation technique 191

(Martin and Bateson, 2007). We observed the dominant male for 15 minutes. Every 15 seconds, 192

we recorded the activity of the dominant male (rest, feed, stand, and walk) as well as his mating 193

behaviours. Behavioural frequencies were divided by the focal duration to estimate the 194

proportion of time spent performing an activity. Focals on the dominant male started when he 195

was active (i.e. not resting) and were not performed more frequently than one focal per hour. We 196

tried to observe every dominant male each day, but only males with the highest status were able 197

to remain dominant in a group throughout the mating season. Dominant males, independently of 198

their age, were observed and the data analysed. Subdominant satellites males were also observed, 199

but the corresponding data was not analysed or included in this study.

200

We summed the proportion of time dominant males spent in particular mating behaviours to 201

define four groups of behaviours representing four tactics : Agonistic corresponds to inter-male 202

competition through agonistic behaviours (Display, Spar, Fight, Displace, Chase); Herd 203

corresponds to male attempt to control female movements (Herd, Chase females ; see Espmark 204

1964 for description) ; Investigate corresponds to males’ assessment of a female reproductive 205

status and the copulation attempts that may result (it includes Flehmen, Investigate, Sniff, 206

Attempt copulation) ; Court corresponds to males mating behaviours which denote male 207

spending time close to a female seeking her attention in the hope of obtaining her agreement to 208

mate with her (Court, Follow female; see de Vos et al., 1967 and Tennenhouse et al. 2012 for 209

description).

210 211

2.5. Statistical analysis 212

We assessed the influence of the operational sex ratio of a group (OSRgroup) and the time of 213

the rut on time dominant males spent in the mating tactics using, for each tactic taken separately, 214

a generalized additive mixed model (GAMM) fitted with a logistic link function and binomial 215

error structure, weighted by the focal duration, and using males identity as random factor 216

(intercept only). We fitted the effect of OSRgroup as linear and quadratic effect (Tennenhouse et 217

al., 2011), and the time of the rut using a smoothing parameter (k = 4). A smoothing parameter of 218

4 was chosen after visual inspection of the temporal patterns obtained.

219

The frequency of mating behaviour is influenced by the potential for this activity, i.e. the 220

number of encounters with a partner/competitor, and by the propensity for this activity, i.e. the 221

likelihood the dominant male will perform the activity at a given encounter (de Jong et al., 222

2012). We therefore introduced a term to control for the potential of each activity. The potential 223

for Agonistic mating behaviour was defined as the number of competitors in the group, i.e. the 224

number of males minus one; the potential for Herd and Investigate mating behaviours were the 225

number of females in the group; and the potential for Court was the number of females in 226

oestrous in the group, i.e. Oestrous groupi, as males do not court anoestrous females, while they 227

herd and investigate all females. The number of encounters in a group may be non-linearly 228

related to the number of partners or competitors present in the group, so we fitted the term 229

Potential both as linear and quadratic.

230

The age of the dominant male has a strong effect on his behaviour and the timing of his 231

mating effort (see introduction). Consequently, each of the above variables was introduced in the 232

model with an interaction with the age of the dominant male, which is a categorical variable:

233

Young < 3 years old (hereafter “young dominant males”); and Adult > 3 years old (hereafter “old 234

dominant males”). The full model is therefore given by equation 5:

235

Equation 5 236

𝐵𝑒ℎ𝑎𝑣𝑖𝑜𝑢𝑟 = 𝐴𝑔𝑒 + 𝑃𝑜𝑡𝑒𝑛𝑡𝑖𝑎𝑙 + 𝑃𝑜𝑡𝑒𝑛𝑡𝑖𝑎𝑙: 𝐴𝑔𝑒 + 𝑃𝑜𝑡𝑒𝑛𝑡𝑖𝑎𝑙²+ 𝑃𝑜𝑡𝑒𝑛𝑡𝑖𝑎𝑙²: 𝐴𝑔𝑒 + 𝑂𝑆𝑅_{𝑔𝑟𝑜𝑢𝑝}+ 𝑂𝑆𝑅_{𝑔𝑟𝑜𝑢𝑝}: 𝐴𝑔𝑒 + 𝑂𝑆𝑅_{𝑔𝑟𝑜𝑢𝑝}² + 𝑂𝑆𝑅_{𝑔𝑟𝑜𝑢𝑝}² : 𝐴𝑔𝑒 + 𝑇𝑖𝑚𝑒 + 𝑇𝑖𝑚𝑒: 𝐴𝑔𝑒

Where Behaviour is the proportion of time spent in a given mating tactic; Potential is the number 237

of individuals with which the dominant male can interact to perform the mating behaviour; Time 238

is the time of the rut centered on the median mate date; Age is the age of the dominant male 239

(young or adult). Interactions are represented by “:”.

240 241

We adopted an all subset approach (Symonds and Moussalli, 2010), and therefore we fitted 242

all of the simpler models derived from the above full model with some conditions. First, if one 243

variable is fitted as a quadratic term, Age interacts with either both terms (i.e., X:Age + X²:Age) 244

or none (i.e., X+X²+Age). Second, Age always interacts with Time if time is in the equation.

245

Third, Age and Potential are always in the equation. Finally, we tested for both quadratic effect 246

and linear effect for the variables Potential and OSRgroup. We chose the best model according to 247

the corrected Akaïke Information Criterion (AICc). We retained the most parsimonious model 248

among the competing models that differed in AICc by less than 2 (Burnham and Anderson, 249

2002). All statistical analyses were performed using R 3.0.3 (R, 2011).

250

From the proportion of deviance explained by the retained model, we calculated the 251

proportion of the explained deviance which is explained by the variables Potential, the OSRgroup

252

and the Time. To do this, we calculated the ratio of proportion of deviance explained by the 253

retained model without one of these variables (and their interaction with Age) to the proportion 254

of deviance explained by the retained model.

255 256

3. Results 257

3.1. Operational sex ratio (OSR), the timing of mating seasons 258

We recorded 843 calf birth between May 2^nd and August 8^th (82.9% of the females gave birth 259

during that period; others were either slaughtered or did not give birth, Table 1). We excluded 57 260

calf birth date from further analyses as they were likely resulting from the second oestrus 261

(corresponding to fertilization occurring in November or later). The estimated median mating 262

date varied between October 1^st and October 17^th (Table 1). The operational sex ratio at the herd 263

level (OSRherd) varied greatly among years (Fig 1a), and on average OSRherd initially decreased 264

and then increased during the peak-rut week for each year taken separately (Fig 1). The OSRgroup

265

varied greatly (average ± sd = 0.79 ± 0.16) from a female biased situation (minimum OSRgroup = 266

0.289) to a highly male biased situation (maximal OSRgroup = 0.995). We observed 222 267

copulations within the two weeks surrounding the estimated mid-peak rut (Fig 1b). These 268

observations are not totally synchronized with the estimated mid-peak rut, as copulations were 269

observed, on average, 1.5 days after the mid-peak rut. This difference is certainly due to a bias in 270

our ability to observe early copulations in the field.

271

272

3.2. Dominant male mating tactics 273

We recorded 1122 focal observations of dominant males, for a total of 276 hours of 274

observation. These records came from the observation of 75 different dominant males (median 275

number of observation per individual = 8). Mating group composition ranged from 1 to 70 276

females (average ± sd = 14.3 ± 11.5 females), and from 1 to 18 males (average ± sd = 2.6 ± 2.7 277

males). We recorded focal observations from 441 young dominant males and 681 adult dominant 278

males. Young dominant males faced competitors in their group in 182 focal observations, while 279

adult dominant males faced competitors in 353 focal observations.

280

Dominant males spent on average 2.5% ± 5.4 of their time performing the mating behaviours 281

analysed in this study, the rest of their time being dedicated to standing, eating, walking and 282

resting. Dominant males spent most of that time in inter-male agonistic behaviours (49.7%), then 283

herding females (26.4%), investigating female reproductive status (15.3%), and courting was the 284

least performed mating tactic (8.4%).

285 286

3.3. Selected models 287

The full model best explained the variability of the time spent in agonistic mating tactics 288

with no competing models. It included the effect of the number of competitors and its quadratic 289

term, the effect of the OSRgroup and its quadratic term, the effect of time, and the interaction of 290

each of these variables with the age of the dominant male (Table 2). The model explained 6.9%

291

of the deviance.

292

The selected model to explain the variability of the time spent herding females was in 293

competition with two other models (ΔAICc = +0.4 for the retained model). It included the effect 294

of the number of females, its quadratic term and their interactions with the age of the dominant 295

male, the effect of the OSRgroup and the effect of time within the rutting season and its interaction 296

with the age of the dominant male (Table 2). The model explained 5.1% of the deviance.

297

The selected model to explain the variability of the time spent investigating females had no 298

competing model. It included the effect of the number of females, the effect of the OSRgroup as 299

quadratic term, the effect of time and the interaction of all of those variables with the age of the 300

dominant male (Table 2). The model explained 8.3% of the deviance.

301

The selected model to best explain the variability of the time spent courting females included 302

the effect of the number of females in oestrous, the effect of the OSRgroup as quadratic term, the 303

effect of time, and the interactions of all of those variables with the age of the dominant male 304

(Table 2) The model explained 6.3% of the deviance.

305 306

3.4. Influence of the potential number of encounters, the OSR and the time 307

For both young and old dominant males, we found the potential number of encounters to 308

have a quadratic relationship with the proportion of time spent in agonistic behaviours (Fig 2a;

309

increasing and then decreasing when more than 9 males are present) and herding (Fig 2b;

310

increasing and then decreasing when more than 22 females are present). As for the time spent 311

investigating females and courting females in oestrous, the relationship with the number of 312

individuals was positive for adult dominant males, but negative for young dominant males (Fig 313

2cd).

314

In general, for both young and adult dominant males, an increase of the competition among 315

males (i.e. increasing OSRgroup) negatively influenced the propensity of males to engage into all 316

mating related behaviours (Fig 3abc). At the highest OSRgroup (OSRgroup > 0.8), however, young 317

dominant males engaged more in agonistic behaviours (Fig 3a), and adult dominant males 318

engaged more in investigating and courting behaviours (Fig 3cd). We observed no influence of 319

OSRgroup on the propensity of young dominant males to engage in courting behaviours (Fig 3d).

320

The different mating tactics were displayed at different time during the rut (Fig 4). Both 321

adult and young dominant males were mostly involved in agonistic behaviours at the end of the 322

peak-rut (Fig 4a). They mostly herded females at the beginning of the peak rut (Fig 4b), and they 323

mostly investigated female reproductive status (Fig 4c) and courted them (Fig 4d) at the end of 324

the peak-rut. The temporal pattern of mating behaviour is less marked for young dominant males 325

than for adult dominant males (Fig 4e).

326

As displayed in Table 3, the potential number of encounters accounted for most of the 327

deviance explained by the inter-male agonistic mating tactic model (62.5%; Table 3). The 328

OSRgroup accounted for a large portion of the deviance explained for the investigating and the 329

courting mating tactics (27.8%, 24.1%, respectively; Table 3). The time within the rutting season 330

accounted for a large part of the deviance explained by the models related to the three female 331

directed mating tactics (Herd 42.8%; Investigate: 33.8%; Court 43.5%; Table 3).

332 333

4. Discussion 334

Our result clearly supported the idea that OSR in ungulates vary throughout the peak rut 335

time, thereby validating the assumption under which we based our predictions. We found indeed 336

that OSR varies for our population both within years, and among years during the study period, 337

being at its smallest values around the mid-peak rutting time. Our results also appeared to show 338

that OSR is not the main predictor of males mating tactics, and that its relation with the 339

propensity of males to engage in mating behaviours is complex.

340

341

4.1 Timing of the rutting season 342

We found that male reindeer clearly displayed a variety of mating tactics, supporting 343

previous reports that most animals (Gross, 1996; Roff, 1996; Oliveira et al., 2008; Neff and 344

Svensson, 2013), including ungulates (Isvaran, 2005; Pintus et al., 2015), are flexible in their 345

mating tactics. More importantly, and in accord with our prediction, we found a sequence in 346

dominant male mating tactics: males were first herding at the beginning of the peak rut week.

347

During the peak rut, dominant males mostly chased other males, as this behaviour is mainly 348

influenced by the number of subdominant males available to chase, which is highest during the 349

peak rut. At the end of the peak-rut, dominant males were mostly investigating and courting 350

females. This sequence appeared to match with a strategy that maximizes access to females in 351

oestrous and thereby optimizing individual reproductive success (Isvaran, 2005; Pintus et al., 352

2015). In a fission-fusion group dynamics system, using a single tactic may not be optimal.

353

Groups are so unstable that harem defense alone is not sufficient, group movements are not 354

spatially predictable and often groups are moving on a too large area to adopt a resource-defense 355

or a lek mating tactics. Moreover, females’ oestrus can be so synchronous that a tending mating 356

tactic would secure too few females. Males herd females before the peak rut to ensure they 357

control a large enough mating group during the peak rut. Also, males tend to defend mating 358

groups during the peak-rut, when herding is less required – as enlarging groups at the end of the 359

peak rut is less beneficial, justifying the tendency for group stability to decrease (Body et al., 360

2015). At the end of the peak rut, a harem defense tactic is costly and risky (as the group may 361

split and females in oestrous may occur by chance in the sub-group leaving), and so it is more 362

efficient for males to use a tending tactic, which is more expected when females are spread out or 363

when they form groups too large to be defended (Emlen and Oring, 1977; Clutton-Brock, 1989;

364

Carranza, 2000; Isvaran, 2005). In conclusion, we can state that instead of an array of mating 365

tactics, reindeer males use a sequence of mating tactics: herding, then chasing, and finally 366

tending (investigating and courting). It is to be expected that this sequence is stable across years, 367

as it will increase male mating opportunities independently of the males-females ratio. Such a 368

sequence of mating tactics seems appropriate for fission-fusion group dynamics systems. Indeed, 369

alternative mating tactics are selected to maximize fitness, leading to the suggestion that such 370

plasticity in mating tactics might represent the adaptive adjustment of the males’ behaviours to 371

differences in social and environmental conditions (Emlen and Oring, 1977; Clutton-Brock, 372

1989; Carranza, 2000).

373

374

4.2 Male ability to perform mating behaviours 375

Our study showed that both young and adult dominant males displayed the above mentioned 376

sequence of mating behaviours. Most discussions of alternative mating tactics in ungulates have 377

looked at populations with a mixed male age structure within a group, most of them showing that 378

adult males tend to monopolize females while younger males usually adopt sneaking tactics 379

(Roed et al., 2002; Willisch et al., 2012; Pintus et al., 2015). Here we show that young dominant 380

males also display mating behaviours often attributed to adult males, such as herding, and in the 381

similar sequence. Alternative mating tactics are therefore a second choice for young males, and 382

they will display harem-defense and tending mating behaviours if given the opportunity.

383

However, we noted some discrepancies in how young and adult dominant males performed 384

them.

385

Both young and adult males display a limit to their herding ability. Males start decreasing 386

their time spent herding when there are more than 22 females to control. Herding is so costly for 387

males reindeer that it may be uneconomical to keep herding while competing with other males at 388

the same time (Brown, 1964; Tennenhouse et al., 2011). Young and adult dominant males 389

herding behaviour therefore do not differ in their propensity to engage into this behaviour, but 390

rather in their timing, young males being unable to match it at the beginning of the peak rut, and 391

to its outcome. Moreover, young males are not efficient at herding females back to the group 392

surely due to their inexperience. Earlier studies in this population suggested already adult 393

dominant males to be more efficient in herding females, and holding larger and more stable 394

mating groups (Holand et al., 2006; Tennenhouse et al., 2011; L'Italien et al., 2012; Body et al., 395

2014).

396

Males also display a limit to their propensity to engage into inter-male agonistic behaviour, 397

and this limit is influenced by their age. Adult dominant males spent less time chasing other 398

males when they were more than 9 other males in the group, while this limit is dropped to 4 other 399

males for young dominant males. There is also a strong difference between adult and young 400

dominant males in their interactions with females: as expected, adult dominant males spent more 401

time investigating and courting females when there were more females in oestrous, as compared 402

to young dominant males. These results are in agreement with other finding, showing that many 403

aspects of male reproduction, such as duration of male-male aggression (Jennings et al., 2004) 404

and copulatory success (Apollonio et al., 1992) are affected by experience.

405

The sequence of mating tactics is also less pronounced for young dominant males than for 406

adult dominant males, mostly for herding and courting behaviours. There is evidence that large 407

males can time their reproductive effort to coincide more precisely with female ovulation than 408

small males (Preston et al., 2003; Meise et al., 2014). Adult male savannah baboons (Papio 409

cynocephalus) appear to compete more intensely for females on the two most likely days of 410

conception (Bercovitch, 1988). All these may again be attributed to experience, and it is clear 411

that adult dominant males are more efficient in timing their reproductive effort (e.g. adult 412

dominant males only spent a small proportion of time investigating) in order to achieve higher 413

reproductive success as compared to young dominant males (Willisch and Ingold, 2007; Willisch 414

and Neuhaus, 2009; Tennenhouse et al., 2012; Willisch et al., 2012; Pintus et al., 2015).

415

416

5. Conclusions 417

Here we have shown that OSR varies through the rut, because of the number of female in 418

oestrous changing with time. We also reported that the level of competition, as measured by the 419

OSR, is not the main driver of male mating behaviours. To monopolize more females in 420

oestrous, dominant males adjust their mating behaviours in relation to the time of the rut, and the 421

social environment. It clearly appeared indeed that young and adult dominant males performed 422

the same ritual when it comes to mating behaviours, following the same sequence: herding, 423

agonistic, investigating and courting. Adult males were however more efficient in timing their 424

effort and performing these mating behaviours than young males, which may explain their ability 425

to monopolize most oestrous female. Our study confirms that reindeer mating strategy is highly 426

flexible, and points to a more complex relationship between mating behaviours and mating 427

success, suggesting that intrasexual variation in mating tactics in relation to time may be 428

adaptive. It also improves our understanding of the mechanism through which dominant males 429

achieve higher reproductive success.

430

431

Acknowledgements 432

The authors thank Jukka Siitari of the Finnish Institute of Natural Resources for the management 433

of GPS collars data, and Mika Tervonen of the Finnish Reindeer Herder’s Association for the 434

management of reindeers in Finland. We thank Sacha Engelhardt, Natalka Melnycky, Hallvard 435

Gjøstein and many others who helped with data collection. We also acknowledge the financial 436

support of the Natural Sciences and Engineering Research Council of Canada (RBW) and the 437

Research Council of Norway (ØH).

438

References 439

Apollonio, M., Festa-Bianchet, M., Mari, F., Mattioli, S., Sarno, B., 1992. To lek or not to lek:

440

mating strategies of male fallow deer. Behavioral Ecology 3, 25-31.

441

Bercovitch, F.B., 1988. Coalitions, cooperation and reproductive tactics among adult male 442

baboons. Animal Behaviour 36, 1198-1209.

443

Body, G., Weladji, R.B., Holand, O., Nieminen, M., 2015. Fission-fusion group dynamics in 444

reindeer reveal an increase of cohesiveness at the beginning of the peak rut. Acta 445

Ethologica 18, 101-110.

446

Body, G., Weladji, R.B., Holand, Ø., Nieminen, M., 2014. Highly Competitive Reindeer Males 447

Control Female Behavior during the Rut. PLoS ONE 9, 1-9.

448

Brockmann, H.J., 2001. The evolution of alternative strategies and tactics, Advances in the Study 449

of Behavior, Vol 30, Academic Press Inc, San Diego, pp. 1-51.

450

Brown, J.L., 1964. The evolution of diversity in avian territorial systems. Wilson Bulletin 76, 451

160-169.

452

Burnham, K.P., Anderson, D.R., 2002. Model selection and multimodel inference: a practical 453

information-theoretic approach. 2nd Edition ed. Springer-Verlag, New York, New York, 454

USA.

455

Carranza, J., 2000. Environmental effects on the evolution of mating systems in endotherms., In:

456

Apollonio, M., Festa-Bianchet, M., Mainardi, D. (Eds.), Vertebrate Mating Systems, World 457

Scientific, pp. 106-139.

458

Clutton-Brock, T.H., 1989. Mammalian mating systems. Proceedings of the Royal Society B:

459

Biological Sciences 236, 339-372.

460

Coltman, D.W., Bowen, W.D., Wright, J.M., 1998. Male mating success in an aquatically mating 461

pinniped, the harbour seal (Phoca vitulina), assessed by microsatellite DNA markers.

462

Molecular Ecology 7, 627-638.

463

Darwin, C., 1871. The descent of Man, and Selection in Relation to Sex. John Murray, London.

464

de Jong, K., Forsgren, E., Sandvik, H., Amundsen, T., 2012. Measuring mating competition 465

correctly: available evidence supports operational sex ratio theory. Behavioral Ecology 23, 466

1170-1177.

467

de Vos, A., Brokx, P., Geist, V., 1967. A review of social behavior of the North American 468

Cervids during the reproductive period. American Midland Naturalist 77, 390-417.

469

Eloranta, E., Nieminen, M., 1986. Calving of the experimental reindeer herd in Kaamanen during 470

1970 - 85. Rangifer, 115-121.

471

Emlen, S.T., Oring, L.W., 1977. Ecology, sexual selection, and the evolution of mating systems.

472

Science 197, 215-223.

473

Espmark, Y., 1964. Rutting behaviour in reindeer (Rangifer tarandus). Animal Behaviour 12, 474

159-163.

475

Gross, M.R., 1996. Alternative reproductive strategies and tactics: diversity within sexes. Trends 476

in Ecology & Evolution 11, 92-98.

477

Hirotani, A., 1989. Social relationships of reindeer Rangifer tarandus during rut - implications 478

for female choice. Applied Animal Behaviour Science 24, 183-202.

479

Holand, O., Weladji, R.B., Roed, K., Gjostein, H., Kumpula, J., Gaillard, J.M., Smith, M.E., 480

Nieminen, M., 2006. Male age structure influences females' mass change during rut in a 481

polygynous ungulate: the reindeer (Rangifer tarandus). Behavioral Ecology and 482

Sociobiology 59, 682-688.

483

Isvaran, K., 2005. Variation in male mating behaviour within ungulate populations: patterns and 484

processes. Current Science 89, 1192-1199.

485

Jennings, D.J., Gammell, M.P., Carlin, C.M., Hayden, T.J., 2004. Effect of body weight, antler 486

length, resource value and experience on fight duration and intensity in fallow deer.

487

Animal Behaviour 68, 213-221.

488

Klug, H., Heuschele, J., Jennions, M.D., Kokko, H., 2010. The mismeasurement of sexual 489

selection. Journal of Evolutionary Biology 23, 447-462.

490

Kojola, I., 1986. Rutting behaviour in an enclosured group of wild forest reindeer (Rangifer 491

tarandus fennicus Lönnb.). Rangifer 6, 173-179.

492

L'Italien, L., Weladji, R.B., Holand, O., Roed, K.H., Nieminen, M., Cote, S.D., 2012. Mating 493

Group Size and Stability in Reindeer Rangifer tarandus: The Effects of Male 494

Characteristics, Sex Ratio and Male Age Structure. Ethology 118, 783-792.

495

Martin, P., Bateson, P., 2007. Measuring behaviour: An introductory guide. Third ed. Cambridge 496

University Press, Cambridge, UK.

497

Meise, K., Piedrahita, P., Kruger, O., Trillmich, F., 2014. Being on time: size-dependent 498

attendance patterns affect male reproductive success. Animal Behaviour 93, 77-86.

499

Mysterud, A., Bonenfant, C., Loe, L.E., Langvatn, R., Yoccoz, N.G., Stenseth, N.C., 2008. The 500

timing of male reproductive effort relative to female ovulation in a capital breeder. Journal 501

of Animal Ecology 77, 469-477.

502

Mysterud, A., Langvatn, R., Stenseth, N.C., 2004. Patterns of reproductive effort in male 503

ungulates. Journal of Zoology 264, 209-215.

504

Mysterud, A., Roed, K.H., Holand, O., Yoccoz, N.G., Nieminen, M., 2009. Age-related gestation 505

length adjustment in a large iteroparous mammal at northern latitude. Journal of Animal 506

Ecology 78, 1002-1006.

507

Neff, B.D., Svensson, E.I., 2013. Polyandry and alternative mating tactics. Philosophical 508

Transactions of the Royal Society B-Biological Sciences 368.

509

Oliveira, R.F., Taborsky, M., Brockmann, H.J., 2008. Alternative reproductive tactics: an 510

integrative approach. Cambridge University Press, Cambridge, UK.

511

Pelletier, F., 2005. Foraging time of rutting bighorn rams varies with individual behavior, not 512

mating tactic. Behavioral Ecology 16, 280-285.

513

Pelletier, F., Festa-Bianchet, M., 2006. Sexual selection and social rank in bighorn rams. Animal 514

Behaviour 71, 649-655.

515

Pintus, E., Uccheddu, S., Roed, K.H., Perez Gonzalez, J., Carranza, J., Nieminen, M., Holand, 516

O., 2015. Flexible mating tactics and associated reproductive effort during the rutting 517

season in male reindeer (Rangifer tarandus, L. 1758). Current Zoology 61, 802-810.

518

Preston, B.T., Stevenson, I.R., Wilson, K., 2003. Soay rams target reproductive activity towards 519

promiscuous females' optimal insemination period. Proceedings of the Royal Society of 520

London Series B-Biological Sciences 270, 2073-2078.

521

R, 2011. R: A language and nvironment for statisticsl computing, R, R Foundation for Statistical 522

Computing, Vienna, Austria.

523

Roed, K.H., Holand, O., Smith, M.E., Gjostein, H., Kumpula, J., Nieminen, M., 2002.

524

Reproductive success in reindeer males in a herd with varying sex ratio. Molecular 525

Ecology 11, 1239-1243.

526

Roff, D.A., 1996. The evolution of threshold traits in animals. Quarterly Review of Biology 71, 527

3-35.

528

Ropstad, E., 2000. Reproduction in female reindeer. Animal Reproduction Science 60-61, 561- 529

570.

530

Skogland, T., 1989. Comparative social organization of wild reindeer in relation to food, mates 531

and predator avoidance. Advances in Ethology 29, 1-77.

532

Symonds, M.R.E., Moussalli, A., 2010. A brief guide to model selection, multimodel inference 533

and model averaging in behavioural ecology using Akaike's Information Criterion.

534

Behavioural Ecology and Sociobiology 65, 13-21.

535

Tennenhouse, E.M., Weladji, R.B., Holand, O., Nieminen, M., 2012. Timing of reproductive 536

effort differs between young and old dominant male reindeer. Annales Zoologici Fennici 537

49, 152-160.

538

Tennenhouse, E.M., Weladji, R.B., Holand, O., Roed, K.H., Nieminen, M., 2011. Mating group 539

composition influences somatic costs and activity in rutting dominant male reindeer 540

(Rangifer tarandus). Behavioral Ecology and Sociobiology 65, 287-295.

541

Uccheddu, S., Body, G., Weladji, R.B., Holand, O., Nieminen, M., 2015. Foraging competition 542

in larger groups overrides harassment avoidance benefits in female reindeer (Rangifer 543

tarandus). Oecologia 179, 711-718 544

Willisch, C.S., Biebach, I., Koller, U., Bucher, T., Marreros, N., Ryser-Degiorgis, M.-P., Keller, 545

L.F., Neuhaus, P., 2012. Male reproductive pattern in a polygynous ungulate with a slow 546

life-history: the role of age, social status and alternative mating tactics. Evolutionary 547

Ecology 26.

548

Willisch, C.S., Ingold, P., 2007. Feeding or resting? The strategy of rutting male Alpine chamois.

549

Ethology 113, 97-104.

550

Willisch, C.S., Neuhaus, P., 2009. Alternative mating tactics and their impact on survival in adult 551

male alpine ibex (Capra ibex ibex). Journal of Mammalogy 90, 1421-1430.

552 553 554

Tables 555

Table 1. Herds compositions (number of females, number of males, and male age structure), calf 556

production (number of calf produced from September or October mating in the herd, and in 557

bracket the calves produced from mating occurring late and therefore excluded from the 558

analysis), the estimated mid-peak rut date (median mate date) and sampling effort (number of 559

focals) 560

Year Enclosure Females Males Males age

Calves (excluded)

Peak-rut Sampling effort

1996 Lauluvaara 46 6 Young 27 (3) 14^th Oct 119

1997 Lauluvaara 47 5 Young 37 (7) 13^th Oct 63

1997 Sinioivi 47 18 Mixed 38 (6) 9^th Oct 70

1999 Sinioivi 75 3 Adult 48 (10) 16^th Oct 107

2000 Sinioivi 74 3 Young 53 (9) 17^th Oct 67

2001 Sinioivi 79 11 Young 63 (4) 7^th Oct 47

2002 Sinioivi 92 4 Mixed 81 (4) 2^sd Oct 72

2003 Sinioivi 52 4 Mixed 44 (4) 8^th Oct 104

2004 Sinioivi 48 5 Mixed 44 (0) 5^th Oct 51

2005 Sinioivi 55 17 Mixed 39 (2) 6^th Oct 64

2006 Sinioivi 80 19 Mixed 67 (1) 1^st Oct 84

2007 Sinioivi 87 24 Mixed 70 (4) 6^th Oct 83

2008 Sinioivi 41 12 Mixed 31 (1) 1^st Oct 57

2009 Sinioivi 42 17 Mixed 39 (0) 1^st Oct 16

2010 Sinioivi 75 24 Mixed 59 (0) 1^st Oct 59

2011 Sinioivi 34 11 Mixed 23 (0) 1^st Oct 59

561 562

Table 2. Model selection based on AIC to explain the variability of the four mating tactics (agonistic, herd, investigate females, court). We 563

present all the models within ΔAICc ≤ 2 or the two models with the lowest AIC if there were only one model within ΔAICc ≤ 2. Bold terms 564

correspond to selected models. An “:” means “interaction”. The age of the dominant male and the potential were always included, and the 565

interaction between time and age was always included if the time variable was included in the model 566

Model Age Potential Potential² Potential : Age OSR OSR² OSR : Age Time : Age AICc ΔAICc Agonistic

1 x x x x x x x x 3040.7 0

2 x x x x x x x 3049.0 8.3

Herd

1 x x x x x x x 1946.3 0

2 x x x x x 1946.7 0.4

3 x x x x x x 1947.6 1.3

Investigate

1 x x x x x x x 1372.1 0

2 x x x x x x x x 1374.6 2.57

Court

1 x x x x x x x 1374.9 0

2 x x x x x x x x 1378.9 4.08

567

Table 3. Proportion (in percent) of the deviance explained by selected models for each mating 568

tactic and proportion (in percent) of that explained deviance which can only be explained by 569

the potential number of encounters, the OSRgroup or the time, with their interaction with the 570

age of the dominant male if included in the model 571

Deviance explained by selected models

Proportion of deviance only explained by

Mating tactics Potential OSRgroup Time

Agonistic 20.5 62.5 11.8 12.6

Herd 9.99 5.0 4.0 42.8

Investigate 12.6 34.2 27.8 33.8

Court 17.8 32.3 24.1 43.5

572 573

Figures captions 574

575

Figure 1. Variation of (a) the herds’ operational sex ratio, and (b) the distribution of the 576

observed copulations throughout the rut. Each year is centered on their estimated median 577

mating date (time = day 0) based on the backdating procedure, and the shaded bar 578

corresponds to the peak-rut week. In (a), solid lines are Lauluvaara herds and dashed lines are 579

Sinioivi herds. The color of the line is proportional to the year of study (darkest = 1996;

580

lightest = 2011) 581

Figure 2. Influence of the potential number of encounters on the proportion of time spent in 582

each mating tactics by young (left panels) and adult (right panels) dominant males. The 583

potential number of encounters correspond to the number of competitors in the group for (a) 584

the inter male agonistic mating tactic, the number of females in the group for (b) the herding 585

mating tactic, and for (c) the investigating mating tactic, and it corresponds to the number of 586

females in oestrous for (d) the courting mating tactic. Partial effect (solid line) and their 95%

587

confident intervals (grey area) were calculated using the median OSRgroup (OSRgroup = 0.48) 588

and at October 1^st (time = 0). Dots correspond to partial residuals averaged (a) per 589

competitor, (b,c) per 5 females, and (d) per 0.25 females in oestrous. Dot sizes are 590

proportional to the number of data. Top and diagonal numbers on each panel indicate the 591

actual value of the matching point which is outside the display range of the y axis 592

Figure 3. Influence of the operational sex ratio in the group (OSRgroup) on the proportion of 593

time spent in each mating tactics (a: inter male agonistic mating tactic; b: herding mating 594

tactic; c: investigating mating tactic; d: courting mating tactic) by young (left panels) and 595

adult (right panels) dominant males. Partial effect (solid line) and their 95% confident 596

intervals (grey area) were calculated using the median potential number of encounters per age 597

class (Competitor: 1/1; Females: 9/13; Females in oestrous: 0.31/0.48; for young/adult 598

dominant males) and at October 1^st (time = 0). The dots correspond to partial residuals 599

averaged per 0.05 unit of OSRgroup. Dot sizes are proportional to the number of data. Top and 600

diagonal numbers on each panel indicate the actual value of the matching point which is 601

outside the display range of the y axis 602

Figure 4. Influence of the time of the rut (centered on the peak rut date: time = 0) on the 603

proportion of time spent in each mating tactics (a: inter male agonistic mating tactic; b:

604

herding mating tactic; c: investigating mating tactic; d: courting mating tactic) by young (left 605

panels) and adult (right panels) dominant males. Partial effect (solid line) and their 95%

606

confident intervals (grey area) were calculated using the median potential number of 607

encounters per age class (see Fig 2), and the median OSRgroup (see Fig 3). The dots 608

correspond to partial residuals averaged per day. Dot sizes are proportional to the number of 609

data. Top and diagonal numbers on each panel indicate the actual value of the matching point 610

which is outside the display range of the y axis. To best compare the timing of each mating 611

tactics, we display (e) the scaled variation of the predictions made on each mating tactic: inter 612

male agonistic behaviour (black solid line), herding behaviour (black dotted line), 613

investigating behaviour (grey solid line), courting behaviour (grey dashed line). The pink bars 614

correspond to the peak-rut week 615

616

Figure 1 617

618

619 620

Figure 2 621

622

623 624

Figure 3 625

626

627 628

Figure 4 629

630

631